Matrix for one page shows up on about:blank page

[gorhill]

Moved from #185:

I have tried going to about:blank to see maybe it needs to be allowed or something, well nothing shows up on the matrix; however when you click on the matrix you find that the last website used is still in the "buffer" if you will, so what's going on? (see picture)

So that in itself is weird to say the least but the fact that access to about:blank by the other extensions is being so aggressively blocked without the ability to whitelist it or even without any explicit settings that interfere with behind-the-scene actions. So what's happening there? (see example)

[gorhill]

I tried to reproduce this and I couldn't. However I am pretty sure I saw this behavior in the past. I will investigate if I can get to happen again.

[GuardianMajor]

Ok, thank you. For me its reproducible by simply going to ANY website, observing the matrix shows things relevant to that site and then type about:blank in the address bar, it loads and click on the matrix and it shows the permissions for the last page you were on before going to it. At least that's how it always happens on my end.

[ghost]

Somehow this reminds me of http://www.wilderssecurity.com/showpost.php?p=2332868&postcount=383

[gorhill]

type about:blank in the address bar, it loads and click on the matrix and it shows the permissions for the last page you were on before going to it

Ok, reproduced. I see. HTTPSB doesn't care about anything else than URLs starting with http://, httpsb:// or chrome-extension://, so anything else won't result in an internal rebinding of the tab with whatever URL ignored by HTTPSB, and this mean the old binding is still looked up.

Somehow this reminds me of http://www.wilderssecurity.com/showpost.php?p=2332868&postcount=383

For this bug here to happen user needs to navigate to a non-http URL (i.e. about:blank, chrome://settings, etc.) in the same tab where a http URL exists. In your linked thread, I see it is about navigating from one http URL to another http, so not related to this one here (still, wondering what could have caused that).

[GuardianMajor]

Yeah I don't think tlu's link is related in this particular case but it does sound interesting that its happening.

As for not caring for non protocol links, can we have a default failsafe that says if non-protocol link then clear matrix? Should be a pretty easy clean up, no?

[gorhill]

It's fixed locally on my computer. I just can't check in now because of other stuff I am working on.

[GuardianMajor]

I took a look at the code and that should do it. So whenever you are done with what you are working on, then we can just push it out in the next package release.

[gorhill]

Fixed with 8db73a1.

[GuardianMajor]

Sweet. I will keep you posted if anything else comes up.